How to increase ARPU in a large B2C business: the five levers that actually move average revenue per user, and the operating model behind them.
Every large B2C business we work with wants to know how to increase ARPU. Almost none of them have a clean answer to what, exactly, is doing the work when ARPU moves. That gap is the whole subject of this piece: the five levers that actually raise average revenue per user, why only two of them compound, and the operating model you need to run them at the same time without cannibalizing each other.
We will get to concrete tactics. But first, a shared definition: if you want the pure textbook version, our guide to ARPU (definition, formula and how to calculate it) covers it in eight minutes. This post assumes you already know the arithmetic and are here for the operational answer.
There are, empirically, five things you can do to increase ARPU in a B2C business. They are not equal. Three of them are one-time gains, two of them compound.
The single most common mistake we see is ranking ARPU opportunities by response rate, revenue, or predicted uplift on the target metric. All three are wrong. The metric that matters is expected incremental margin per action, which is a very specific quantity:
Expected incremental margin equals the probability that the customer takes the action because of your intervention, multiplied by the margin on what they take, minus the cost of the intervention itself. It rewards actions that would not have happened without you and penalizes those that would have. It is the only ranking that survives a look from the CFO.
A plan upgrade that a customer would have made in three weeks anyway is worth almost nothing when you nudge it today. A save offered to someone who was never going to churn is a discount given away. A cross-sell that just moves revenue between your own products is worse than doing nothing. Every ARPU program needs to see and rank on incrementality, or it silently pays for value it did not create.
Sustainable ARPU growth is not a campaign. It is a rate. The organizations that raise ARPU quarter after quarter share three properties, none of them optional.
Not just monthly. Not just for the customers due for renewal this week. Every customer, every day, gets an updated propensity and uplift score for each of the five levers. That is the raw material an agentic loop or a mature analytics team can rank against. Without per-customer per-lever scores, ARPU interventions default to segments, and segment-level treatments systematically undershoot per-customer uplift.
Every eligible action has a per-customer expected margin, net of intervention cost. The workspace picks the top-ranked action per customer per moment, subject to channel, frequency and eligibility constraints. This is exactly the operating logic of Next Best Action, applied specifically to ARPU-relevant interventions.
A permanent 5 to 15 percent slice of the eligible base sees no interventions from the ARPU program. That is your control. Every weekly readout is treated-versus-holdout ARPU delta by lever, cohort and product. This is the number the CFO trusts because it is the number seasonality and mix cannot fake.
A large share of what looks like retention or upsell success in legacy CRM programs is measurement artifact. The customers most likely to accept a discount are usually the ones most likely to have stayed anyway. The customers most likely to upgrade in response to a promo are usually the ones about to upgrade organically. When you measure lift as pre/post or as year-over-year on treated customers alone, the artifact looks like a program win. Against a holdout, it disappears.
The practical implication is: your incremental ARPU lift is smaller than your program dashboard says, and probably a lot smaller. The way out is not to shame the CRM team. It is to install the holdout and start ranking actions by incremental margin so the program earns real gains it can defend.
The five-lever framework is universal, but the mix that compounds fastest depends on the vertical. A few examples we see repeat:
If none of this is running yet, the fastest path to visible incremental ARPU is narrower than most programs assume. Pick one lever (usually plan mix), one product, one channel, and a preserved holdout. Ship the ranked actions weekly against the holdout for six weeks. Read the incremental margin. Then widen. Everything downstream of that first loop, from measurement infrastructure to team roles to reviewer discipline, is described in our playbook on continuous decisioning.
Markin runs this loop against your base 24/7. To see the same five levers ranked live for your customers, explore Growth optimization.
Frequently asked
Related resources
Solutions
See it in the product
The same loops this note describes run 24/7 against your customer base. Watch the workspace decide, experiment and execute 1:1.
